Network-Aware Virtual Workload Placement in Data Centers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional cloud computing systems fail to optimally deploy virtual workloads due to their inability to account for network resource states, leading to suboptimal performance and inefficient utilization of computing resources across multiple data centers.
Innovation Solution
A virtual workload management component that considers both computing resource hardware load and network hardware load, using network state data and topology information to select the most suitable physical servers for workload deployment and perform live migration to optimize network resource utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If virtual workloads are deployed without considering network resource states, then deployment speed is improved, but workload performance deteriorates
Solution Approach 1:
The system performs preliminary evaluation of network resource states and computing resource states before deploying virtual workloads. The workload management component assesses network hardware load, network state data, and topology information in advance to identify optimal deployment locations, ensuring both fast deployment and high performance by pre-determining suitable physical servers based on current system conditions.
2Device complexity
If workloads are statically allocated to physical servers, then system complexity is reduced, but resource utilization efficiency deteriorates
Solution Approach 1:
The system implements dynamic workload allocation where the workload management component continuously monitors network resource states, computing resource hardware load, and network state data. Based on real-time conditions, the system automatically migrates virtual workloads between physical servers to optimize resource utilization. This dynamic approach balances management complexity with improved productivity by adapting to changing system conditions.
3Adaptability or versatility
If multiple data centers are used to provide computing resources, then system versatility is improved, but optimal workload placement becomes more difficult
Solution Approach 1:
The workload management component implements a universal evaluation framework that handles multiple data centers with diverse computing resources and network configurations through a single standardized interface. The system evaluates network resource states, computing resource hardware load, and network state data using consistent criteria across all data centers, enabling optimal workload placement regardless of the specific data center environment or resource type.
Data Source
AI summary
Techniques for virtual workload deployment based on computing resource hardware load and associated network hardware load. For each of a plurality of computing resources within one or more data centers onto which a virtual workload can be deployed, a computing resource hardware load of the respective computing resource is determined. Network topology information is maintained for at least one network fabric of the one or more data centers, and an associated network hardware load of a network device communicatively connected to the respective computing resource is determined. Embodiments automatically select one or more computing resources. The virtual workload is deployed onto the automatically selected one or more computing resources.